to-UTF-16.pl

   1 #! /usr/bin/perl
   2
   3 use strict;
   4 use warnings;
   5
   6 my($in_name, $out_name) = @ARGV;
   7
   8 open my $in_h, "<", $in_name or die "$in_name: $!";
   9 binmode $in_h or die "$in_name: $!";
  10 my $head;
  11 defined(read $in_h, $head, 4) or die "$in_name: $!";
  12 close $in_h or die "$in_name: $!";
  13
  14 my $charset;
  15 if ($head =~ /^\x00\x00\xFF\xFE/) {
  16         print "$in_name: Found big-endian UTF-32 BOM\n";
  17         $charset = "UTF-32";
  18 } elsif ($head =~ /^\xFE\xFF\x00\x00/) {
  19         print "$in_name: Found little-endian UTF-32 BOM\n";
  20         $charset = "UTF-32";
  21 } elsif ($head =~ /^\xFE\xFF/) {
  22         print "$in_name: Found big-endian UTF-16 BOM\n";
  23         $charset = "UTF-16";
  24 } elsif ($head =~ /^\xFF\xFE/) {
  25         print "$in_name: Found little-endian UTF-16 BOM\n";
  26         $charset = "UTF-16";
  27 } elsif ($head =~ /^\xEF\xBB\xBF/) {
  28         print "$in_name: Found UTF-8 BOM\n";
  29         $charset = "UTF-8";
  30 } else {
  31         print "$in_name: No BOM found -- assuming UTF-8\n";
  32         $charset = "UTF-8";
  33 }
  34
  35 open STDOUT, ">", $out_name or die "$out_name: $!";
  36 exec "iconv", "-f", $charset, "-t", "UTF-16", $in_name
  37         or die "exec iconv: $!";